What the image compressor does
The compressor re-encodes an image at a quality level you choose, in JPG, PNG or WebP, and shows the size before and after so you can judge the trade-off. It uses the browser's own image encoder through the Canvas API, which means the picture is decoded, redrawn and re-saved on your device — nothing is uploaded. Typical results: a 4 MB phone photo becomes 400–800 KB at 80% JPEG quality with no visible difference at normal viewing size.
How lossy compression works
JPEG and lossy WebP discard information the eye is bad at noticing: fine colour detail (the eye resolves brightness far better than hue), and high-frequency texture in busy areas. The encoder splits the image into small blocks, converts each to frequency components, and rounds the least-visible components toward zero. The quality slider controls how aggressive that rounding is. Because the discarded data is gone for good, re-compressing an already compressed image loses a little more each time — always compress from the best original you have.
| Quality | Typical size vs original JPEG | Visible effect |
|---|---|---|
| 90–100% | 70–120% | None; often larger than the source at 100% |
| 80% | 40–60% | None at normal size — the sweet spot |
| 70% | 30–45% | Slight softening in fine texture |
| 50% | 15–30% | Blockiness in gradients, halos around edges |
| 30% and below | under 15% | Obvious artefacts; suitable only for thumbnails |
Choosing an output format
- JPG — photographs and anything with gradients. Universally supported; no transparency.
- WebP — 25–35% smaller than JPG at equal quality, supports transparency, and is accepted by every current browser. Some older desktop software and email clients still cannot open it.
- PNG — lossless, so the quality setting has no effect. Choose it for screenshots, logos, text and line art, where JPEG halos would be visible. It is the wrong format for photos: a photo saved as PNG is often 5–10× larger than as JPG.
- Same as original — keeps the format and just re-encodes at the new quality.
Why compression alone is sometimes not enough
File size depends on pixel count as much as on quality. A 4,000 × 3,000 photo has 12 million pixels; a web page or chat app displays it at 1,200 pixels wide at most. Resizing to 1,600 px on the long edge cuts the pixel count by 85% before any quality reduction, and the result still looks sharp on a phone or laptop screen. For the smallest files, resize first with the Image Resizer, then compress. If you need to hit an exact limit (a 200 KB upload cap, say), use Compress to Target Size, which searches for the quality that lands under the limit.
What happens to metadata
Redrawing through a canvas strips EXIF data: camera model, exposure settings, date taken and GPS location are all removed from the output. For sharing publicly that is usually desirable — location data in particular is something most people do not intend to publish. If you need to keep the metadata (for photography portfolios or archiving), use the JPG Compressor (Keep EXIF) tool, which copies the original EXIF block into the compressed file.
Common uses
- Shrinking photos for email or messaging without visible loss.
- Preparing product and blog images for the web, where page weight affects search ranking and load time.
- Meeting upload size limits on job sites, government portals and marketplaces.
- Reducing a folder of phone photos before backing up to limited cloud storage.
- Converting PNG screenshots that were saved as photos into properly sized JPGs.